Transcript
Page 1: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Energy Efficient Mobile Applications with Mobile Cloud Computing (MCC)

Dr BHOJAN ANAND

Page 2: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ School of Computing, National Uni. of Singapore

The ‘Garden City’

Page 3: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ School of Computing, National Uni. of Singapore

106 years old 7,000 staff 33,000 students

Page 4: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Let's Make this Interactive

Please feel free to interrupt and share your thought.

Page 5: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+In this talk

n Gamelet - Distributed Cloud Infrastructure

n Elpincel - Cloud support for Adaptive Display Power Management

Page 6: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile Computing Changes Lifes

Price comparison Entertainment Mobile search

Navigation Learning

Page 7: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Smartphones are Intimate

n Mobile  phones  and  millennials  (gen  Y)    (Cisco  2012):  

•  75%  use  in  bed  before  sleep    •  83%  sleep  with  their  phones  •  90%  check  first  thing  in  the  morning  

•  A  third  use  in  bathroom  (!!)  •  A  fiHh  check  every  ten  minutes  

Page 8: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile Devices in Use

n Over 90% in most parts of the world n Some parts are 100+%

(Singapore. 152%) n 73% India (23% of which are

smartphones)

n Web Access: 70% of rural India's active Internet population access the web via mobile phones (25 million users in rural India)

Source: TRAI

Page 9: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Opportunitis in mobile computing…

n What do users want?

n How to meet their needs?

Page 10: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Quantity (Bandwidth) vs Functionalities n If a quantity is changed by 10 times, there will be

qualitative changes in functionalities.

n GSM Data 9.6 Kbps: audio

n GPRS 100 Kbps: video QVGA 320x240

n 3G 2 Mbps: video VGA 640x480

n 4G/LTE 12 Mbps (75 Mbps Theoritical): HD Video

n 100 Mbps: Interactive Virtual Reality?

n Creativity has no limit. Users always want more.

Page 11: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Users always want more - Display

n Display n Smartphones with five-inch screens or bigger

rose 369% in the first three months of this year n LCD -> OLED (vivid colors and sharp) n  Flexible or Bendable displays

n à Smartwatch

Page 12: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Users always want more – Processors, Sensors n Quad core Processors n Render HD media, 3D Games

n 64-bit Architecture

n More sensors with greater accuracy and new form connectivities n STEP Sensor – makes indoor

localisation feasible accross multiple floors

Page 13: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Users always want more –

n  Navigation systems goes beyond simple GPS

n  A floor plan overlays on top of the physical floor

n  Multi-player shared games will require pose/gesture recognition and emotion sensing, and rich graphics

n  Interactive Virtual Reality environments for Mobile

n  Phones learns user bahaviours (higly personalised) --- Adaptive OS

n  Integrated Authentication and Payment System

n  ….. and more

AR Map (iPhone)

Page 14: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Users always want more –

n These Applications Relies on …. n speech recognition, natural language

processing, computer vision and graphics, machine learning, augmented reality, planning and decision making.

n What it is common among the above functions?

n Solution?

Page 15: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile Cloud Computing

n Small form factor mobile device

+

n Nearly unlimited computing resources of Cloud

Mobile Cloud Computing (MCC) refers to set of techniques that use near unlimted cloud

resources to empower mobile applications.

What is seacial about it? How it differs from Cloud Computing?

Page 16: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Common Arctitecture Approch for MCC

Offloading Computation and Storage

Page 17: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Benefits of using Clould services for Mobile Computing n Enhance computational and

storage capability

n Energy efficiency of the mobile device

n Opportunity to gather more sensing data (such as encountering other mobile devices with different sensing abilities)

n Access to different content/data sets that are for eg. location specific (such as social net data)

Page 18: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+MCC - Some Key Challenges: n Processing Latency vs Communication Latency

n Even with LTE, access to the closest data center will incur a latency of at least 70ms

n Processing Energy vs Communication Energy n Offloading computation is not always effective in

saving energy

n Always-on connectivity, On-demand Bandwidth scalability n One of the biggest issues, because the radio resource

for wireless networks is much more scarce than wired networks

n Security and Privacy

Page 19: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Alternate Architectures are required –Local Clould n Moore’s law is expected to stay for at least

another 5-10 years. n A Buck for Billion. “Billion Transistors on a Chip”

n What is means?

n Cloud Systems will move out of Huge Data Centers n @ Telco towers n @ ISPs n Or even Mobile Peers

Page 20: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Approaches for Arctitecture of MCC - Local Peer-Cloud

Attractive Incentives Security and Privacy

May be interstying for groupof users working on a same problem!

Page 21: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Approaches for Arctitecture of MCC – Local Micro-Cloud

A case for Micro-Cloud and an Application follows …

Page 22: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Mobile Cloud Computing eg. Gamelets Target Hardware: CPU/GPU

Target Application:Real-time Mobile Multiplayer Games

Bhojan Anand, Aw Jia Hao Edwin , "Gamelets - Multiplayer Mobile Games with Distributed Micro-Clouds," [appeared IEEE-ICMU 2014]

Page 23: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ Real-time Multiplayer Games

Page 24: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Real-time Multiplayer Game Charateristics n Mobile Games and Game Maps

n Characteristics (FPS games) n Highly interactive, Closed maps with occlusions, 40-80

fps

n ~20 pps server to client (in

regular interval, cannot burst!)

n ~40 pps client to server (in regular interval, cannot burst!)

n Latency more than 180ms makes the game play annoying and not playable

Page 25: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Real-time Multiplayer Game Charateristics n Characteristics (MMORPG games) n Slow speed games, with few high interactive areas n Huge (100’s of square miles) open maps with few

occlusions. [Estimated – time taken to run from one end to another].

n 100s to few 1000s of players per server

n 40-60 fps n ~15 pps server to client n ~20 pps client to server n Latency more than 400ms

makes the game play annoying and not playable

Page 26: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Real-time Multiplayer Game Charateristics

How did conventional multiplayer games handle latency issue?

Page 27: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Real-time Multiplayer Game Charateristics n How did conventional multiplayer games

handle latency issue?

Local Game State Simulation

Immediate Response

Prediction Local

Perception Filters

Require Intelligent

Clients!

Page 28: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

28

Event Collection

Client Frame (60 fps)

Server Frame (20 fps)

Player/Opponent Extrapolation and

Interpolation

Render Graphics & Audio

Game Simulation (Mechanics,

Physics, AI, etc) Extrapolation to

Reduce BW

Game State

Event Collection

Client Frame (60 fps)

Player/Opponent Extrapolation and

Interpolation

Render Graphics & Audio

Game Action

Game Action

State Sync

State Sync

Local Game State

Local Game State

Latency Hiding and Synchronisation

Page 29: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Multiplayer Cloud Games

n Latency Hiding Techniques

n But, what happens when the client is dump?

n as in current CLOUD games

Immediate Response Prediction

Local Perception

Filters

Require Intelligent

Clients!

Page 30: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ Mobile Multiplayer Cloud Games

•  CPU/GPU load is reduced to save energy

•  Centralized control

•  Scalable

•  Portable

Page 31: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile Multiplayer Cloud Games Key Challenges

-  All these operations should happen serially in few ms!

-  Requires

high sever bandwidth for Video Streaming

Inte

ract

ion

Del

ay

Page 32: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Multiplayer Cloud Games

n “the company had deployed thousands of servers that were sitting unused, and only ever had 1,600 concurrent users of the service worldwide.”

http://www.theverge.com/2012/8/28/3274739/onlive-report

Active cloud gaming providers G-cluster Gaikai (subsidiary of Sony Computer Entertainment) GFACE Kalydo OnLive Playcast Media Systems StreamMyGame NVIDIA GRID

Page 33: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Multiplayer Cloud Games

n Could Games introduce additional latencies n Cloud Processing n Encoding Video

n More serious one is: n Network Latency and Jitter n Worse in Mobile Environments. n Even with LTE it takes 70ms to closest data center n Goes upto 200ms

Key Challenge

Page 34: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Gamelet System

n  Gamelet is a minimal-set hardware required to run, render and stream 3D games placed in the same local network or few hops (most of the time one or two hops) away from the mobile client.

n Uses distributed rendering to counter the limitations of renderer hw resources

Render at next hop!

Page 35: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Gamelet Architecture

Benefits:- -  Latency Hiding

(Immediate Feedback, DR, LPF)

-  Scalable (Bandwidth) -  Distributed Rendering

COMPONENTS:- - Processing Unit, - RAM , - Graphics Card, - Flash Drive for basic OS

Peer device can be a Gamelet!

Page 36: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Gamelet System

n Zone Distribution

n Distributed Rendering (Peer-Assisted Rendering)

n  Content Based Adaptive Streaming

Focused Areas of this Work

Page 37: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Zone Distribution & Distributed Rendering n Avg Size of a 3D Game 5 Gbytes

n Eg. Battlefield 3 recommends 4GB RAM and 20GB Storage

n It will take about 56 mins to download over 3G network (assuming 1.5 Mbps)

Download/Render only the “Zones of

Interest”

Check adjacent Gamelets before

downloading Zones

Page 38: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Zone Distribution Zones and What they Contain

Page 39: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Zone Subdivision Algorithm

Ø Max Zone Size= function(t_Download, t_Load)

Ø  Recursively divide until the zone size is less than max size

Ø Dynamically Resized

Page 40: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Zone Request Processing

n Request to download n When user (player

camera) enters the boundary

n  Request to load n  When user’s far clip

plane enters a new zone

Boundary Size = function(far clip plane distance, t_Download, t_Movement)

Page 41: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Distributed Rendering

n Common Methods/Libraries n Network-Integrated Multimedia Middleware

(NMM) n Top Game engines do not have NMM layer

n Real-Time Scene Graph (RTSG) n Not accessible for Game developers

n  Multi-Camera Distributed Rendering (MCDR) n Practicable Approach n Can be used with any Game Engine

Page 42: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+MCDR - Rotating Camera

70 mm è 35mm + 35mm

Page 43: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+MCDR - Rotating Camera

Fish eye view on objects near and far

NOT RECOMMENDED!

Page 44: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Page 45: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+MCDR - Reshaping the view frustum n Most Appropriate Way

-  Settings are Same as Main camera.

-  Manipulate Side Clip Planes

Page 46: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Distributed Rendering Rendering sections of view port in parallel

Page 47: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Selecting Adjacent Gamelets

n Initial adjacency list is obtained from the main Game Server

n List is maintained with periodic ‘pings’ to neighbors

n A Gamelet will send data to neighbor node only if …. n Send Data n Render n Receive Image

In Serial < 40ms (1/25 s)

Page 48: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Streaming to Client

n 30% of the display is used for in-game HUD n 5-8fps is sufficient for such static areas n We mask this area in all other frames, resulting in

high compression ratio

Content Based Adaptive Streaming (CBAS)

Page 49: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Streaming to Client

n When the player is idle and his view is static, the background is mostly static n Stream at low rate

Content Based Adaptive Streaming (CBAS)

Page 50: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Game Client

n The client side code of the game simply gets the compressed stream from the Gamelet and uncompresses it to display.

n  In addition, it captures all the user actions.

n [No Audio in Current Version]

Display and Collect User Actions

Page 51: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Evaluations

n  800x480 WVGA

n  1 to 3 Mbps after state-of-the-art compression with CBAS

n Support upto 5 clients in 802.11b

n 9.6 Kbps between Gamelet to Server

n Server Scalability Increases by at least 100 times (1000/9.6)

Bandwidth & Scalability

Page 52: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Evaluations

n  Measured with windows task manager and MSI Afterburner version 2.3.1

n GPU Load is proportional to number of pixels to render. [Linear]

Average Processing Load

Page 53: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Evaluation

n Garden of Eden – 3D survival Game

n Game Server in School Network

n Gamelets and Clients at two randomly selected 802.11 APs in School

n Laptop, iPad versions of the Client were used

n Seven UG users played the Game. Played multiple versions after 20mins training (Game mechanics) n  Observed artefacts (latency, jitter, visual quality,

synchronisation etc)

Small Scale User Study

Page 54: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Evaluation Small Scale User Study

Page 55: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Contribution

n First attempt towards a distributed micro-cloud infrastructure

n Multi-camera Distributed Rendering n Can be done on top of any game engine

n Content Based Adaptive Streaming Techniques for Games

n Basic Results are Promising

Page 56: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Demo

n  Look for “Gamelets - Multiplayer Mobile Games with Distributed Micro-Clouds” in Youtube

Page 57: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Key Take Away …

n As processing and storage cost reduces n Cloud systems will be available very close to

client

n Current CLOUD games assume Dump Clients n No game knowledge. No latency hiding

techniques n Wireless latency is still very high

n Solution è Gamelets

Page 58: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Open Issues

n  Classical Argument - Distributed vs Central?

n  How handle Security and Cheating issues?

n  How do we know the adjacent Gamelets? Who maintains the list?

n  What happens when the user moves from the area of one Gamelet to another?

n  When to distribute the rendering task and to how many Gamelets?

n  Who hosts the Gamelets?

n  Many mobile clients have good processing power. Can we exploit this?

n  ----- many more …..

Page 59: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Mobile Cloud Computing Cloud support for Adaptive Display Power Management

Page 60: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Limitations of Mobile Computing

Weight Size

Performance Energy

Page 61: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Well Known Problem n Demand and Supply

HTC Desire HD

8h 10 min, 420h

5h 20 min, 490h

1h 50 min

Page 62: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile phone power consumption

n Mobile phone subscriptions worldwide has surpassed 6 billion

n Annual electricity consumption for a mobile phone 11 KWh per year.

n Mobile devices: ~13% of global ICT energy.

Including data centre energy consumption!

Ref: Bhojan Anand ‘Energy efficient algorithms and techniques for wireless mobile clients’, NUS Scholar Bank

Page 63: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Where does the energy go?

Bhojan Anand et al., ‘Adaptive Display Power Management for Mobile Games’, Mobisys 2011

HOW DID WE MEASURE?

Page 64: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Observations by other researchers

©: Corell.

Locally Stored Video

Downloading a 15 MillionBits file via HTTP using wget

Page 65: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Energy Efficiency in Display & Network subsystems

n Applications with Cloud Support can also help in improving the energy efficiency of Display and Network in mobile devices

n Will see 1 example work on Display Energy ….

Page 66: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Mobile Cloud Computing - eg. Elpincel - HVS based Color Transformation Target Hardware: OLED Displays and and its Variants such as AMOLED

Target Application: Images, Texts, Videos, Games

-  Bhojan Anand , et. al, "El-pincel: a painter cloud service for greener Web Pages," ACM Multimedia 2012

-  Bhojan Anand, et.al, "PARVAI - HVS Aware Adaptive Display Power Management for Mobile Games," [to appear,IEEE- ICMU- 2014]

Page 67: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Why Organic LED (OLED)? n Alternate display technology to LCD. (Growing faster!)

Samsung Panasonic Motorola Nokia HTC LG > 20 models 102p RAZR i Lumia820 One-S Optimus

5.5” note II

Larger Screens

7.7” Galaxy Tab

Much wider use – TVs, Cameras, etc

55”

OLE

D T

V

Page 68: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Why bother?

* So

urce

s: E

ngad

get

& L

G D

isp

lay

Page 69: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Why bother?

n Popular websites uses White or Light color background

n 65% of the web contents are White

Page 70: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Organic LED (OLED)

Brighter contents consumes more power

Each pixel is individually illuminated in OLEDs

Factor 1: Brightness / Luminance

700 mW

255 mW

Page 71: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Organic LED (OLED)

n Disparity in OLED sub-pixel power consumption

Factor 2: Color / Hue

two sub-pixels per pixel

RGBG - Pentile

n Eg. Blue OLED material n Lowest luminance efficiency

(lu/Watt)

n Higher current is used to match luminance with Green

n Results in Lower Lifetime

n Solution: Larger Blue Sub-pixel. (Variation in Size & Current for each sub-pixel)

Page 72: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Basic Solution 1: Web Design

ADOBE - NORMAL

ADOBE - MOBILE ADOBE - OLED

Page 73: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Basic Solution 1: Web Design

n Myth-> Mobile = Less n ‘the mobile use case’ has changed - Long

sessions are common

n 50% of the webpages visited by iPhone users are Non-Mobile Versions – Study at RICE

univ

n è W3C recommends – “One Web” n Usable Web for All devices

Page 74: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Basic Solution 2 Existing Work n  OLED Power Consumption = f

(pixel brightness, pixel color) n Bright Pixels consume more

power

n P(Blue) > P(Red) > P(Green) or P(Blue) > P(Green) > P(Red)

n Existing Work: n Make Text/Background Dark or

Green (or May be Red)

n Make Images Dark

41%

Page 75: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ Basic Solution 2 Existing Work 1. Poor Chromatic and

Achromatic Contrast 2. Brand identity is Lost 3. Pixels obtained from

Browser’s Frame buffer to transform (Unnecessary Complexity! and Penalty!)

4. Works only for custom made browser

5. Naïve Image Transformation

41%

Page 76: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Our Solution

n Transform before rendering (Cloud Service) n  Browser Independent n  Client side cost = 0

n Retain Brand Identity

n Ensure Chromatic and Achromatic contrast at recommended level

n Quality – User Defined (0 – 60% saving)

n HVS based image processing

60%

20%

Page 77: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Introducing… El-pincel Cloud Service

n Repaints web pages to energy efficient colors

n No processing cost to client

n Modifies: n Text / Layout n Images

Page 78: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+System Components

Page 79: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Client Analyzer

n Communicates with client

n Client communicates via “agent”

n Details Gathered n Device Model, Screen Type, User

Requirements

Only need to configure the Browser

Page 80: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+System Components

Page 81: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Brand Color Manager

n Obtain key image (default:- Favicon)

Page 82: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Brand Color Manager

n Extracts dominant colors

n Ranks color according to presence

n Pick m most power efficient colors for target device

Page 83: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+System Components

Page 84: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Color Palette Generator

n Input: Cascading Style Sheets (CSS) file

n From the m colors, select a color (base color) which matches the user’s energy efficiency requirement.

n Set of Monochromatic/Analogous colors (equal energy) – Background

Harmonic Color Wheel Types

Page 85: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Color Palette Generator

n Complementary colors – Foreground (e.g. text)

n  Additional Constraint: W3C recommended Chromatic and Achromatic contrast

n Output: New CSS generated

Harmonic Foreground Colors

Page 86: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Color Palette Generator

n Energy Efficiency Matching (Approx.)

19% text pixels

13% text pixels

Page 87: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+System Components

Page 88: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ Image Tone Mapping Engine

n Based on properties of Human Visual System (HVS)

n Human eye is more sensitive towards green color and red than blue n  STEP 1: Energy Efficient Color

Mapping

n Human eye is more sensitive to local contrast compared to global contrast n  STEP 2: Local Contrast Based

Darkening

Our approach

Page 89: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Energy Efficient Color Mapping

For each color C = {r, g, b},

Map to a new color C’= {r’,g’,b’} with minimum power consumption

With constraints: - Hue change - Saturation change - Constant brightness/luminance

Optimization problem

Page 90: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Local Contrast Based Darkening

n Darkens image based on local contrast n Area with edges depicts structure

Page 91: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Local Contrast Based Darkening

- =

Higher intensity of darkening in area with low contrast

Local contrast mask

Page 92: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Local Contrast Based Darkening

Higher intensity of darkening in area with low contrast

Page 93: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Image transformation 60% saving

Relative darkening

Simple darkening

Original

Our approach

Page 94: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Page 95: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

Page 96: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Image transformation 60% saving

Relative darkening

Simple darkening

Original

Our approach

Page 97: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Image transformation 60% saving

Relative darkening

Simple darkening

Original

Our approach

Page 98: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Web page transformation

Original Modified

20% saving

http://www.apple.com/

Page 99: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Web page transformation

Original Modified

60% saving

http://www.adobe.com/

Page 100: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+ Demo

n  Look for “El-pincel : A Painter Cloud Service for Greener Web Pages” in Youtube

Page 101: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Follow-up works n The idea is already embrased by Industry

(Samsung Devices)

Page 102: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Two Key Areas to Focus

n Compute Intensive Applications -> Mobile Could Computing n Eg. Gamelet

n Smaller Devices (and Higher Quality Demand) -> Application Aware Energy Efficiency suppoterd by Cloud Infrastructure n Eg. Elpincel

Page 103: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+More details at …. n  Bhojan Anand, et. al. "Adaptive Display Power Management for Mobile Games”, ACM

MobiSys 2011, Washington DC, June 2011. (http://doi.acm.org/10.1145/1999995.2000002) n  Bhojan Anand , et al. "El-pincel: a painter cloud service for greener Web Pages,” ACM MM

2012. (http://dl.acm.org/citation.cfm?id=2000037) n  Bhojan Anand, Aw Jia Hao Edwin , "Gamelets - Multiplayer Mobile Games with Distributed

Micro-Clouds," IPSJ-IEEE ICMU 2014 (Search in IEEE Xplore during Second Week of Jan ) n  Bhojan Anand, et al. "PARVAI - HVS Aware Adaptive Display Power Management for Mobile

Games," IPSJ-IEEE ICMU 2014 (Search in IEEE Xplore during Second Week of Jan ) n  Bhojan Anand, et al."Game Action Based Power Management for Multiplayer Online

Game," ACM MobiHeld 2009.(http://dl.acm.org/citation.cfm?id=1592619)

Page 104: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+Demos and Tools at ….

n www.comp.nus.edu.sg/~bhojan

n For tools / Source codes: n Email to: [email protected]

Page 105: Energy Efficient Mobile Applications with Mobile Cloud Computing ( MCC )

+

THANK YOU Questions?


Recommended